Nonlinear compensation method and system for rotary direct-drive servo valve based on neural network

Abstract

The present application belongs to the technical field of pre-stage driving error compensation of rotary direct drive servo valve, and provides a nonlinear compensation method for rotary direct drive servo valve based on neural network to solve the problem of various nonlinear errors of rotary servo valve in practical application. The method comprises the following steps: obtaining the operating parameters of the rotary direct drive servo valve, then constructing a nonlinear compensation neural network, determining the preset target displacement of the valve core of the rotary direct drive servo valve, inputting the preset target displacement of the valve core into the nonlinear compensation neural network, obtaining the corresponding compensation current, converting the compensation current into error compensation torque, transmitting the error compensation torque to the controller of the rotary direct drive servo valve, and adjusting the displacement of the valve core according to the error compensation torque, so as to compensate for the various nonlinearities existing in the system caused by the slide valve reversing hysteresis, slide valve negative opening dead zone, motor rotation friction and the like, and improve the reliability and stability of the system.

Description

Technical Field

[0001] This invention belongs to the field of rotary direct-drive servo valve pre-stage drive error compensation technology, specifically relating to a nonlinear compensation method and system for rotary direct-drive servo valves based on neural networks. Background Technology

[0002] A finite-angle motor-driven rotary servo valve is a type of valve that utilizes an electric motor for precise control of fluid flow and pressure. It converts the position of the motor into the valve's opening degree, allowing for precise flow regulation. These servo valves are commonly found in industrial automation and hydraulic systems, characterized by fast response and high control accuracy.

[0003] In rotary servo valves driven by finite-angle motors, various nonlinear errors exist due to factors such as spool valve hysteresis, spool valve negative opening dead zone, and motor rotation friction, causing deviations between the actual output torque and the target torque. By utilizing the self-learning capability and arbitrary approximation characteristics of neural networks, nonlinear compensation can be effectively performed, thereby improving the accuracy and reliability of control. Through training, neural networks can identify and correct these nonlinear characteristics, achieving more precise dynamic control. Summary of the Invention

[0004] In order to solve at least one of the above-mentioned technical problems in the prior art, the present invention provides a nonlinear compensation method and system for rotary direct drive servo valves based on neural networks.

[0005] According to the first aspect, a nonlinear compensation method for a rotary direct-drive servo valve based on a neural network includes the following steps:

[0006] Obtain the operating parameters of the rotary direct drive servo valve and establish a dynamic model of the rotary direct drive servo valve. The operating parameters include at least: actual valve core displacement, armature current, motor rotation angle, and motor speed.

[0007] A nonlinear compensation neural network is established based on the operating parameters. The nonlinear compensation neural network includes an input layer, a hidden layer, and an output layer. The input layer includes five nodes, which correspond to the actual displacement of the valve core, the armature current, the motor rotation angle, the motor speed, and the preset target displacement of the valve core, respectively. The hidden layer includes six nodes, which correspond to nonlinear sign functions, respectively. The output layer obtains the compensation current of the rotary direct drive servo valve based on the processing results of the input layer and the hidden layer.

[0008] The preset target displacement of the valve core of the rotary direct drive servo valve is determined, and the preset target displacement of the valve core is input into the nonlinear compensation neural network to obtain the corresponding compensation current. At the same time, the error compensation torque is determined based on the compensation current and the dynamic model of the rotary direct drive servo valve.

[0009] The error compensation torque is transmitted to the controller of the rotary direct drive servo valve, and the valve core displacement is adjusted according to the error compensation torque.

[0010] Preferably, the dynamic model is as follows:

[0011]

[0012] In the formula, x represents the valve core displacement parameter; v represents the valve core rotation speed parameter; T and T f These represent the torque and torque disturbance, which are proportional to the control current, respectively; f1 is a linear function with motor parameters as variables, and f2 is a nonlinear function affected by nonlinear errors; t represents time.

[0013] Preferably, after establishing the nonlinear compensation neural network based on the operating parameters, the method further includes:

[0014] Obtain the historical operating dataset of the rotary direct drive servo valve. The historical operating dataset includes at least one data sample, which is the historical operating parameters and the historical compensation current obtained after inputting the historical operating parameters.

[0015] The nonlinear compensation neural network is trained based on the historical running dataset to obtain a nonlinear compensation neural network with convergent parameters.

[0016] Preferably, training the nonlinear compensation neural network based on the historical running dataset includes:

[0017] The historical valve core target displacement and historical operating parameters in the historical operating dataset are input into the nonlinear compensation neural network to obtain the corresponding historical compensation current;

[0018] Based on the historical compensation current and the dynamic model, the historical error compensation torque is determined;

[0019] The valve core movement of the rotary direct drive servo valve is controlled based on the historical error compensation torque, and the historical valve core error displacement is determined according to the historical valve core target displacement.

[0020] The historical valve core displacement error is fed back to the nonlinear compensation neural network, and the weight values ​​of the nonlinear compensation neural network are adjusted.

[0021] The historical valve core error displacement is used as the new input to the nonlinear compensation neural network model after weight adjustment. The weight values ​​are adjusted until the error between the historical valve core target displacement and the historical valve core displacement error is lower than the preset threshold. The model parameters of the nonlinear compensation neural network model at this time are saved to obtain the nonlinear compensation neural network with converged parameters.

[0022] Preferably, the weight values ​​of the nonlinear compensation neural network are adjusted using the following formula:

[0023]

[0024] In the formula, x(k) is the valve core displacement. The output of the model training is Y(j) = [y(1)...y(n)]. T It is the neuron input vector, and

[0025] W(i,j)=[w(i,1)w(i,2)...w(i,n)] T These are the weight coefficients during k-1 training iterations, where n is the number of input variables in the input layer, and i is the number of neurons in the layer.

[0026] Preferably, it further includes:

[0027] The gradient of each neuron in the output layer of the nonlinear compensation neural network model is calculated using the chain rule and propagated layer by layer until a preset number of iterations is reached.

[0028] According to the second aspect, a nonlinear compensation system for a rotary direct-drive servo valve based on a neural network is capable of performing the method described in the first aspect and any preferred embodiment, comprising: a controller, a magnetostrictive displacement sensor, a current sensor, and an angular displacement sensor.

[0029] The controller is used to receive the error compensation torque and control the movement of the valve core according to the error compensation torque;

[0030] The magnetostrictive displacement sensor is used to provide a valve core displacement signal, which includes the actual valve core displacement and the error valve core displacement.

[0031] The current sensor is used to measure armature current;

[0032] The angular displacement sensor is used to measure the motor rotation angle and the motor speed.

[0033] Compared with the prior art, the beneficial effects of the present invention are:

[0034] This invention can compensate for many nonlinearities in the system caused by spool valve hysteresis, spool valve negative opening dead zone, and motor rotation friction without changing the existing rotary direct drive servo valve pre-stage drive hardware, thereby improving the system's reliability and stability. Attached Figure Description

[0035] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0036] Figure 1 This is a schematic diagram of a nonlinear compensation method for a rotary direct-drive servo valve based on a neural network, provided in an embodiment of the present invention.

[0037] Figure 2 This is a schematic diagram of the workflow of the nonlinear compensation neural network provided in an embodiment of the present invention. Detailed Implementation

[0038] The technical solutions of the embodiments of the present invention will be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other implementation methods obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0039] It should be noted that the structures, proportions, sizes, etc., shown in the accompanying drawings of this specification are only for the purpose of assisting those skilled in the art in understanding and reading the content disclosed in the specification, and are not intended to limit the conditions under which the present invention can be implemented. Therefore, they have no substantial technical significance. Any modifications to the structure, changes in the proportional relationships, or adjustments to the size, without affecting the effects and objectives that the present invention can produce, should fall within the scope of the technical content disclosed in the present invention. It should be noted that in this specification, relational terms such as "first" and "second" are only used to distinguish one entity from several other entities, and do not necessarily require or imply any actual relationship or order between these entities.

[0040] Rotary direct-drive servo valves are high-performance hydraulic control components. They are hydraulic control valves that receive analog electrical signals and output modulated flow and pressure accordingly. They offer advantages such as fast response and high control precision, and are widely used in aerospace, robotics, and other fields. However, traditional rotary direct-drive servo valves employ complex structures such as spool valves, nozzle-baffle valves, and jet pipe valves. In practical applications, factors such as spool valve reversing hysteresis, spool valve negative opening dead zone, and motor rotation friction cause various nonlinear errors in the system, resulting in deviations between the actual output torque and the target torque. To address these issues, this invention provides a neural network-based nonlinear compensation method and system for rotary direct-drive servo valves. This method promptly adjusts the nonlinear errors generated during application, improving overall stability.

[0041] This invention provides a nonlinear compensation system for a rotary direct-drive servo valve based on a neural network, comprising: a controller, a magnetostrictive displacement sensor, a current sensor, and an angular displacement sensor;

[0042] The controller is used to receive the error compensation torque and control the movement of the valve core according to the error compensation torque; the magnetostrictive displacement sensor is used to provide the valve core displacement signal, which includes the actual valve core displacement and the error valve core displacement; the current sensor is used to measure the armature current; and the angular displacement sensor is used to measure the motor rotation angle and the motor speed.

[0043] like Figure 1 As shown, a flowchart illustrating a neural network-based nonlinear compensation method for a rotary direct-drive servo valve, applied to the neural network-based nonlinear compensation system provided in the above embodiments, is presented, including the following steps:

[0044] S1: Obtain the operating parameters of the rotary direct drive servo valve and establish a dynamic model of the rotary direct drive servo valve. The operating parameters include at least: actual valve core displacement, armature current, motor rotation angle, and motor speed.

[0045] In this embodiment, the dynamic model is as follows:

[0046]

[0047] In the formula, x represents the valve core displacement parameter; v represents the valve core rotation speed parameter; T and T f These represent the torque and torque disturbance, which are proportional to the control current, respectively; f1 is a linear function with motor parameters as variables, and f2 is a nonlinear function affected by nonlinear errors; t represents time.

[0048] S2: A nonlinear compensation neural network is established based on the operating parameters. The nonlinear compensation neural network includes an input layer, a hidden layer, and an output layer. The input layer includes five nodes, which correspond to the actual displacement of the valve core, the armature current, the motor rotation angle, the motor speed, and the preset target displacement of the valve core, respectively. The hidden layer includes six nodes, which correspond to nonlinear sign functions, respectively. The output layer obtains the compensation current of the rotary direct drive servo valve based on the processing results of the input layer and the hidden layer.

[0049] In this embodiment, the nonlinear compensation neural network is set with an input layer, including five nodes, which correspond to the preset target displacement of the valve core, the actual displacement of the valve core, the armature current (feedback from the current sensor), the motor rotation angle (feedback from the angular displacement sensor), and the motor speed, respectively.

[0050] A hidden layer is set up, consisting of six nodes, corresponding to nonlinear symbolic functions.

[0051] Set up an output layer to output the compensation current required for the error compensation torque.

[0052] Optionally, after establishing the nonlinear compensation neural network based on the operating parameters, the method further includes: obtaining a historical operating dataset of the rotary direct drive servo valve, wherein the historical operating dataset includes at least one data sample, the data sample being the historical operating parameters and the historical compensation current obtained after inputting the historical operating parameters; and training the nonlinear compensation neural network based on the historical operating dataset to obtain a nonlinear compensation neural network with converged parameters.

[0053] In this embodiment, the training process of the nonlinear compensation neural network can be completed using the MATLAB Neural Network Toolkit. This method can also be used with other tools for training, and this application is not limited to these.

[0054] Optionally, training the nonlinear compensation neural network based on the historical operating dataset includes: inputting the historical valve core target displacement and historical operating parameters from the historical operating dataset into the nonlinear compensation neural network to obtain the corresponding historical compensation current; determining the historical error compensation torque based on the historical compensation current and the dynamic model; controlling the valve core movement of the rotary direct drive servo valve based on the historical error compensation torque, and determining the historical valve core displacement error according to the historical valve core target displacement; feeding the historical valve core displacement error back to the nonlinear compensation neural network and adjusting the weight values ​​of the nonlinear compensation neural network; using the historical valve core displacement error as the input to the nonlinear compensation neural network model after weight adjustment, and continuing to adjust the weight values ​​until the error between the historical valve core target displacement and the historical valve core displacement error is lower than a preset threshold, saving the model parameters of the nonlinear compensation neural network model at this time, and obtaining a nonlinear compensation neural network with converged parameters.

[0055] In this embodiment, the historical valve core target displacement x is input into the nonlinear compensation neural network. d (1) and the corresponding historical actual displacement x(0), and based on the historical operation dataset, give the control torque T(1) corresponding to the incorrect compensation current and the historical valve core error displacement x(1) controlled by the incorrect compensation current. x d The error between (1) and x(1) is used to adjust the weight values ​​of the neural network through the backpropagation process.

[0056] In this embodiment, the weight values ​​of the nonlinear compensation neural network are adjusted using the backpropagation method. The weight values ​​W of the neural network are adjusted to minimize the following error values, as specified in the formula:

[0057]

[0058] In the formula, x(k) is the valve core displacement. The output of the model training is Y(j) = [y(1)...y(n)]. T It is the neuron input vector, and W(i,j)=[w(i,1)w(i,2)...w(i,n)] T These are the weight coefficients during k-1 training iterations, where n is the number of input variables in the input layer, and i is the number of neurons in the layer.

[0059] Adjustments are made based on the calculation results of this formula. When the sum of squared errors is less than 0.001, the optimal weight adjustment result is output.

[0060] The historical valve core error displacement x(1) is used as the new input of the nonlinear compensation neural network model after weight adjustment, generating new parameters x(2), I(2), θ(2), ω(2). The weight values ​​are adjusted according to the new parameters until the error between the historical valve core target displacement and the historical valve core displacement error is lower than the preset threshold. The model parameters of the nonlinear compensation neural network model at this time are saved, and the nonlinear compensation neural network with converged parameters is obtained.

[0061] Optionally, it also includes: calculating the gradient of each neuron in the output layer of the nonlinear compensation neural network model using the chain rule, and propagating it layer by layer until a preset number of iterations is reached.

[0062] In this embodiment, the gradient of the loss with respect to each parameter (weights and biases) is calculated using the chain rule. The nonlinear compensation method includes: calculating the partial derivative of the loss function with respect to the output to obtain the gradient of each neuron in the output layer, and propagating the gradient layer by layer backward. In each layer, the gradient of the current layer is calculated, and the gradient of the previous layer is calculated based on the output of the current layer and the gradient of the previous layer. Using the calculated gradients, the weights and biases are adjusted through an optimization algorithm. The backpropagation process is repeated on each training sample until the network converges or reaches the set number of iterations.

[0063] S3: Determine the preset target displacement of the valve core of the rotary direct drive servo valve, and input the preset target displacement of the valve core into the nonlinear compensation neural network to obtain the corresponding compensation current. At the same time, determine the error compensation torque based on the compensation current and the dynamic model of the rotary direct drive servo valve.

[0064] S4: Transmit the error compensation torque to the controller of the rotary direct drive servo valve, and adjust the valve core displacement according to the error compensation torque.

[0065] In this embodiment, as Figure 2As shown, the preset target displacement of the valve core is input into the trained nonlinear compensation neural network, and the output layer obtains the corresponding compensation current. Then, the controller calculates the error compensation torque based on the compensation current and the dynamic model of the rotary direct drive servo valve, and then controls the movement of the valve core of the rotary direct drive servo valve based on the error compensation torque.

[0066] In summary, compared with the prior art, this invention compensates for many nonlinearities in the system caused by spool valve hysteresis, spool valve negative opening dead zone, and motor rotation friction without changing the existing rotary direct drive servo valve pre-stage drive hardware, thereby improving the reliability and stability of the system.

[0067] The above description is merely a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A nonlinear compensation method for rotary direct-drive servo valves based on neural networks, characterized in that, Includes the following steps: Obtain the operating parameters of the rotary direct drive servo valve and establish a dynamic model of the rotary direct drive servo valve. The operating parameters include at least: actual valve core displacement, armature current, motor rotation angle, and motor speed. The dynamic model is as follows: In the formula, Represents the valve core displacement parameter; This represents the valve core rotation speed parameter; and These represent the torque and torque disturbance, which are proportional to the control current, respectively. It is a linear function with motor parameters as variables. It is a nonlinear function affected by nonlinear errors; Indicates time; A nonlinear compensation neural network is established based on the operating parameters. The nonlinear compensation neural network includes an input layer, a hidden layer, and an output layer. The input layer includes five nodes, which correspond to the actual displacement of the valve core, the armature current, the motor rotation angle, the motor speed, and the preset target displacement of the valve core, respectively. The hidden layer includes six nodes, which correspond to nonlinear sign functions, respectively. The output layer obtains the compensation current of the rotary direct drive servo valve based on the processing results of the input layer and the hidden layer. The preset target displacement of the valve core of the rotary direct drive servo valve is determined, and the preset target displacement of the valve core is input into the nonlinear compensation neural network to obtain the corresponding compensation current. At the same time, the error compensation torque is determined based on the compensation current and the dynamic model of the rotary direct drive servo valve. The error compensation torque is transmitted to the controller of the rotary direct drive servo valve, and the valve core displacement is adjusted according to the error compensation torque.

2. The nonlinear compensation method for a rotary direct-drive servo valve based on a neural network according to claim 1, characterized in that, After establishing the nonlinear compensation neural network based on the operating parameters, the method further includes: Obtain the historical operating dataset of the rotary direct drive servo valve. The historical operating dataset includes at least one data sample, which is the historical operating parameters and the historical compensation current obtained after inputting the historical operating parameters. The nonlinear compensation neural network is trained based on the historical running dataset to obtain a nonlinear compensation neural network with convergent parameters.

3. The nonlinear compensation method for a rotary direct-drive servo valve based on a neural network according to claim 2, characterized in that, The training of the nonlinear compensation neural network based on the historical running dataset includes: The historical valve core target displacement and historical operating parameters in the historical operating dataset are input into the nonlinear compensation neural network to obtain the corresponding historical compensation current; Based on the historical compensation current and the dynamic model, the historical error compensation torque is determined; The valve core movement of the rotary direct drive servo valve is controlled based on the historical error compensation torque, and the historical valve core error displacement is determined according to the historical valve core target displacement. The historical valve core displacement error is fed back to the nonlinear compensation neural network, and the weight values ​​of the nonlinear compensation neural network are adjusted. The historical valve core error displacement is used as the new input to the nonlinear compensation neural network model after weight adjustment. The weight values ​​are adjusted until the error between the historical valve core target displacement and the historical valve core displacement error is lower than the preset threshold. The model parameters of the nonlinear compensation neural network model at this time are saved to obtain the nonlinear compensation neural network with converged parameters.

4. The nonlinear compensation method for a rotary direct-drive servo valve based on a neural network according to claim 3, characterized in that, The weight values ​​of the nonlinear compensation neural network are adjusted using the following formula: In the formula, For valve core displacement, The output of the model training. It is the neuron input vector, and Is Weight coefficients during training. It is the number of input variables in the input layer. It is the number of neurons in a layer.

5. The nonlinear compensation method for a rotary direct-drive servo valve based on a neural network according to claim 3, characterized in that, Also includes: The gradient of each neuron in the output layer of the nonlinear compensation neural network model is calculated using the chain rule and propagated layer by layer until a preset number of iterations is reached.

6. A neural network-based nonlinear compensation system for a rotary direct-drive servo valve, capable of executing the neural network-based nonlinear compensation method for a rotary direct-drive servo valve as described in any one of claims 1-5, characterized in that, include: Controllers, magnetostrictive displacement sensors, current sensors, and angular displacement sensors; The controller is used to receive the error compensation torque and control the movement of the valve core according to the error compensation torque; The magnetostrictive displacement sensor is used to provide a valve core displacement signal, which includes the actual valve core displacement and the error valve core displacement. The current sensor is used to measure armature current; The angular displacement sensor is used to measure the motor rotation angle and the motor speed.

Images